You have been granted access to this page through First Click Free. Subsequent use of TabbFORUM will require logging in. If you don't have an account, registration is free.

Videos

  • Rail_thumb_adam_honore_david_etue-finqloud-safenet-cloud_security_in_financial_services

    Debunking the Cloud Security Myth

    The idea that the cloud is not secure is a misperception, says David Etue, SafeNet. While the cost and agility gains that come with moving to the cloud also come with a loss of some...
     
  • Rail_thumb_steve_phillips-nasdaq_omx-latam

    Latin America's Quest for Liquidity

    Brazil no longer is the only game in town when it comes to trading in Latin America. According to Steve Phillips, senior managing director, Latin America and Caribbean, NASDAQ OMX...
     
  • Rail_thumb_adam_will_screen_shot

    What Will Nasdaq Do With eSpeed?

    Following several failed cross-border exchange mergers, Nasdaq’s acquisition of the electronic fixed income trading platform eSpeed provides the exchange with a way to diversify...
     
 

More Video | Podcasts

Advertisement
Original_eu-13284_300x250-banner_2
Original_broadridge_300x250_multi-asset
Original_equityjune_300x250
Original_cn_mkts300x250b
Spotlight-blackInnovations in Trading and Technology (more stories)

05 October 2010

Low-Latency Networking for Financials

Theodoras focuses on the three major sources of delay in getting between two points in a financial district.

I attended several back-to-back financial trading conferences in New York City earlier this year, with a refreshed perspective on the state of low-latency networking of financial institutions.

The field of low-latency networking for financial players continues to evolve and innovate at a relentless pace. Just a little over a year ago, trades were won or lost by milliseconds. By the end of the year, that was shaved to microseconds. At the conferences, I was told that today 250ns can make the difference between winning and losing a trade. Times have changed, indeed.

With this new business reality comes increased focus on every source of latency in a path between two points. I say "two points," as there are a wide variety of locations and business sites being connected in the financial realm, including but not limited to exchanges, information feeds, co-location providers, hedge funds, broker/dealers, etc. It is no longer sufficient to simply upgrade routers, or buy the latest multi-core processor based computer blades. All sources of latency must be minimized.

The three major sources of delay in getting between two points in a financial district are 1) fiber optic cable, 2) proximity and 3) transport delays.

1)      Fiber optic delay is simply the time the data spends in the fiber optic cable. The shorter the cable, the less the delay. While cables are envisioned to run directly between two locations, the truth is much less ideal. Financial districts are complex rat's nests of fiber cables crisscrossing a city, crossing roads, going up and down manholes, and along available easements. Shortening the network’s fiber length often means finding a dark fiber provider that has pieced together a shorter route.

2)      Proximity delay is how close you can get to the available fiber access points. It is a basic geometric problem. There can only be so much real estate nearby a fiber junction, and locating your equipment in buildings closest to this point lowers the delay, often at a co-location provider’s site.

3)      Transport delay is the amount of latency that is added by any and all optical gear that the data encounters as it is transported along the fiber. In the past, when trade execution times were in the seconds, the transport delay was consider inconsequential. There is a misconception that data travels at the speed of light in a fiber, so there is no real way of speeding it up. In reality, transport delay is a real contributor to path latency, and it can be significantly lowered with a variety of strategies.

The fiber optic link between two financial sites is anything but bare cable. There are several optical transport functions that are needed for one’s data to ride on a fiber: 1) Color conversion, 2) Optical amplification, 3) Dispersion compensation, and 4) Electrical regeneration.

1) The majority of optical transport links today are Wavelength Division Multiplexed (WDM), which means everyone is assigned a color channel. In order to transport one’s data over a WDM network, it must first be converted from grey to a color. This function is called “transponding”. If done incorrectly, milliseconds can be incurred here.

An often overlooked feature of color conversion is “muxponding,” or the aggregation of lower speed traffic into a higher speed signal. For example, today most information feeds are 1Gbit/s, yet most transport links are 10Gbit/s. So, frequently 10 different information feeds are squeezed into a single transport color. The standard way of doing this is ODU encapsulation, which can add double digit microseconds to a link, on each end.

2) A signal traveling down an optical fiber gets smaller and smaller with each kilometer of distance. Optical amplifiers are used to boost the signal as it weakens. Traditionally, a type of optical amplifier called and EDFA (Erbium Doped Fiber Amplifier) is used, and the delay through EDFA’s was considered negligible. However, in today’s trading environment where every nanosecond counts, the hundreds of nanoseconds incurred at each amplifier along a path can no longer be ignored. Some high gain, dual stage EDFA’s can have delays in the low microseconds.

3) Just as rain in the sky, or glass prisms can spread light into a rainbow of colors, so do fiber optic cables. This smearing of an optical data signal into multiple colors is predominately an issue at 10Gbits/s, and can cause a signal’s assigned color to bleed into neighboring channels. This prism effect is called Chromatic Dispersion (CD). Traditionally, long spools of a special type of fiber called Dispersion Compensating Fiber (DCF) has been used to reverse the smearing of colors. However, as these spools can add up to hundreds of additional kilometers of fiber in a path, they are poorly suited to low-latency applications.

My own firm has developed alternative dispersion compensation techniques that remove the need for large spools of DCF – Fiber Bragg Gratings (FBG’s). While a full review of FBG technology is beyond the scope of this article, FBG’s can be thought of as a reverse prism written into a piece of fiber optic cable less than a meter long. They essentially have negligible delay, and greatly reduce the end-to-end latency of a path.

4) Finally, even with the help of optical amplification and dispersion compensation, on longer links a signal will degrade to the point where it needs to be regenerated. How the signal gets regenerated greatly determines the additional path delay incurred. Traditional regeneration techniques can add hundreds of microseconds of unnecessary delay. Now, more than ever, latency is key to financial business models, and every nanosecond counts. This article has but touched on the complex topic of sources of latency in optical transport networks.

To learn more about latency and the wide range of technologies ADVA Optical Networking has developed to speed all optical transport functions, please visit ADVA Optical Networking's low latency landing page herehttp://blog.advaoptical.com/category/fields/low-latency-transport

Spotlight-white-trans For more stories in the Innovations in Trading and Technology Spotlight Series click here.

Comments | Post a Comment

2 Comments to "Low-Latency Networking for Financials":
  • Comment_larry_tabb
    ltabb

    05 October 2010

    its amazing how important such little time increments are. Who ever thought we would be fighting over milliseconds more over, microseconds, or nanoseconds. What are we going to fight over after nanoseconds? If it gets to that point, who will even be able to play? Or will technology become so advanced that a nanosecond will become second nature?

  • Missing
    david.brukman

    06 October 2010

    Just as important as engineering a microsecond- or soon nanosecond- optimized connection, is the validation, monitoring and diagnostic means that work at that level. With almost any component in the chain able to add hundreds of microseconds to the overall latency and dwarf the painstaking optimization elsewhere, the ability to verify performance, detect when it degrades, and pinpoint the cause of it become crucial. For instance, something as previously trivial as technician having problems with a designated port and then, using a different port, different patch path, or a different piece of equipment to bypass the problem, can blow up the connection latency. Without the proper monitoring tools in place, you may not even realize the extra microseconds until your fill ratio drops.

You must log in to comment.